The present invention relates to a method of controlling smooth running such as that known, for example, from German Patent Document DE 195 48 604 C1. The known method is used for determining differences of the torque contributions of individual cylinders of an internal-combustion engine by means of the course of the rotational crankshaft speed. This method is based on a recognition that the rotating movement of the crankshaft takes place in an irregular manner under the effect of gas forces and forces of gravity. In order to determine the rotational-speed fraction or torque fraction of a cylinder, individual cylinders are cut off in a targeted manner during engine operation. By means of a comparison with the rotational speed course of the engine operated without a cylinder cut-off, the torque fraction of each individual cylinder in the overall engine torque can be illustrated separately by means of a rotational speed signal. The injection quantity spreadings caused by manufacturing tolerances are recognized and are to be compensated for by establishing the same average pressures in all cylinders by the variation of injection quantities.
A similar method is described in German Patent Document DE 41 22 139 C2. This method is also based on the fact that cyclic irregularities occur; these cyclic irregularities are caused by the different quantities of fuel injected into the individual cylinders of the internal-combustion engine because of tolerances in the injection devices. The starting point is the fact that the torque or the rotational acceleration is directly proportional to the injected fuel quantity. In order to avoid rotational speed irregularities, the fraction of each combustion process in the rotational acceleration is detected. The measured values are compared with one another by forming average values, and deviations are determined in this manner. The fuel injection quantities of the individual cylinders are finally changed such that the deviations disappear. The sum of the changes of the fuel quantity injected into the individual cylinders is selected such that it results in a total of zero.
In the case of an internal-combustion engine according to International Patent Document WO 97/23716, the fuel supply to a cylinder can be cut off. The cylinder will then operate, for example, as a compressor. In order to avoid vibrations in this method of operation, the fuel supply to the remaining, normally operating cylinders is changed in the appropriate manner. It is possible to determine by experiments and calculation in which manner the torque of the cylinders is to be distributed in order to achieve an optimal suppression of vibrations. For certain operations, determined data are kept available in this manner according to which the internal-combustion engine is controlled. The injection quantities are obviously distributed to the individual cylinders such that the vibrations of the 0.5th to 3rd orders are suppressed because only they are responsible for noticeable vibrations in practice. However, the vibrations of the various orders can obviously not always be suppressed to the same extent. The appropriate fuel distribution is obviously related to the size of the vector which is responsible for the vibrations.
A method for the cylinder-selective control of a compression ignition internal-combustion engine is known also from International Patent Document WO 98/07971. In this case, a measuring device is known for detecting the angle of rotation of the crankshaft and for determining the momentary rotational speed of the crankshaft. From the rotational speed of the crankshaft, a control unit determines suitable parameters which permit, in various operating ranges of the internal-combustion engine, a cylinder-selective equalization or a defined inequalization of the mean pressures, in which case the effects of the component differences of the fuel supply and of the combustion system on the combustion process are minimized.
In a dissertation by Jochen Tonndorf, xe2x80x9cInfluence of the Misfire Operation on the Torsional Vibration Behavior of Driving Systems with Piston Enginesxe2x80x9d, authorized by the Mechanical Engineering Department of the Technical University of Rheinland-Westfalen in Aachen, the torsional vibration behavior of engines is studied. It is stated there that operating conditions exist which differ significantly from the normal operation. Thus, tolerance-caused manufacturing differences in a cylinder and the injection system and also deviations caused by wear in the course of the operating duration lead to differences in comparison to the normal operation. As a result, performance deviations of the individual cylinders of approximately +/xe2x88x9210% can supposedly be caused, which results in generation of a torsional vibration exciting force. In multi-cylinder engines, deviations of the individual cylinders may add up so unfavorably that the effect is the same as that of a complete failure of a cylinder. Furthermore, disturbances in the injection system may result in a misfire operation. Damaged inlet or outlet valves may result in a loss of compression. The cut-off of cylinders also represents an operating instance which changes the torsional vibration strain. The effect of the operating conditions deviating from the normal operation on the excitation behavior of the engine is illustrated by a vector representation of the exciter forces. Furthermore, it is stated that, in the misfire operation, only the exciting forces of the 0.5th, 1st and 1.5th order are of interest. The exciting alternating torque is computed from the vector sum corresponding to the phase position of the harmonic. However, the author reaches the conclusion that interventions at the engine, for example by changing the ignition pressure, cannot be carried out in practice.
It is an object of the invention to illustrate a smooth-running control, particularly for internal-combustion engines with high cylinder numbers.
While, in the case of internal-combustion engines with a few cylinders, the rotational speed fractions resulting from the individual cylinders can clearly be detected in the rotational speed curve of an operating cycle, this is not so in the case of internal-combustion engines with large cylinder numbers. On the contrary, the rotational speed fractions are superimposed such that, when viewing the rotational speed curve, conclusions can no longer be drawn with respect to the provoking cylinder, which requires new analyzing methods. Nevertheless, the inventive method can also be applied to internal-combustion engines with a low number of cylinders, although limitations exist there because of the low number of cylinders. For smooth-running control, the low-frequency vibration fractions are considered here. For this purpose, the pulse response spectrum of each cylinder is determined by calculation or measurement. For determining the pulse fraction of a cylinder from the rotational speed by measuring, the cylinders are individually cut off successively and the rotational speed is recorded above the crank angle. In addition, the rotational speed course of the healthy intact engine, that is, when all cylinders are operating normally, is recorded. This may be a new engine directly from the factory in normal operation which, because of tolerances, has slight differences in the rotational speed fractions of each cylinder, or it may be an ideal engine whose cylinders are equalized, for example, by using the method according to the invention, with respect to their fractions in the rotational speed acceleration.
xe2x80x9cIdealxe2x80x9d in this sense means that, before recording the reference values, an adjustment is carried out, for example, by varying the injection quantities of individual cylinders. During this adjustment, the fluctuations of the rotational speed contributions of the cylinders are minimized. This adjustment is maintained in the normal operation. By forming the difference between the course of curve of the healthy engine and of the courses of the curves for individually cut-off cylinders, new curves are generated which reflect the influence of each cylinder on the overall rotational speed course. These response curves are subjected to a Fourier decomposition. However, only low-frequency harmonic vibrations, expediently of the 0.5th to 3rd order, are considered, and the pertaining spectral pulse responses {right arrow over (I)} of the rotational speed course of an operating cycle of each cylinder are recorded. In the normal engine operation, the rotational speed course of the crankshaft is now entered continuously above the angle, and analogously, by means of a Fourier decomposition of the obtained course of the curve, the spectrum {right arrow over (R)} of an operating cycle is formed. For illustrating the spectral rotational speed course again only the Fourier coefficients of the low-frequency vibrations are used, specifically preferably of the harmonics of the 0.5th to 3rd order which are processed to form a row matrix. The spectral pulse responses {right arrow over (I)} and the resultant {right arrow over (R)} of Fourier coefficients of the rotational speed course can be illustrated for each harmonic as a vector indicator above the crank angle. When the resultant is equal to zero, no correction of the injection quantities is required. However, when a resultant is present, this means that an insufficient injection is taking place in a cylinder, and, as a result of the correction of the injection quantities of the individual injectors, the resultant must be changed to zero. The distribution of the total injection quantity required for the given load case takes place such that the components of the resultant situated in the direction of the pulse response indicator are multiplied by the pulse responses {right arrow over (I)}. The result is correction factors for the injection quantities. Cylinders which are situated in the direction of the resultant {right arrow over (R)} are more corrected by means of positive or negative signs than those situated more orthogonally. The mathematical operation, which can accomplish the corresponding task is the formation of the scalar product or of the vectorial inproduct from the resultant {right arrow over (R)} and the spectral pulse responses {right arrow over (I)}. For this purpose, the required data are held available in matrix form. The matrix multiplication of the pulse responses {right arrow over (I)} with the vector of the spectral rotational speed course {right arrow over (R)} results in values different from zero and leads to a correction of the injection quantities when a smooth running deviation exists in the normal operation. The correction values, which are normalized, are supplied to a governor and the injection quantities xcex94Q are determined, which may be positive or negative, and correspondingly correct the injection quantities for each injector of a cylinder determined by the engine governor.